Boys Calorimeter

Charles Brainin and Atharv Kulkarni

This project was completed as part of the Department of Physics and Astronomy’s Contemporary Physics (PHYS174). This project was advised by Professor Stecklein and Hardy Richardson with additional assistance from Eric Struth and Mark Wilson.

Introduction

Calorimetry is one of the foundational experimental techniques in thermodynamics and thermal physics. By measuring the transfer of heat between systems, calorimeters allow physical quantities such as specific heat capacities, latent heats, and enthalpies of reaction to be experimentally determined. Combustion calorimetry in particular has long been used to characterize the energy content of fuels and remains important in industrial chemistry, power generation, and engineering applications.

In this project, we constructed and tested a low-cost hybrid Boys-flow calorimeter designed to measure the heating value of propane gas through direct thermal transfer into flowing water. The apparatus was inspired by the classic Boys calorimeter, originally developed as a simple method for estimating the calorific value of gaseous fuels by measuring heat absorbed by flowing water surrounding a combustion chamber. Our design additionally incorporates elements of flow calorimetry, allowing continuous water circulation through copper tubing exposed to combustion gases.

A major goal of this project was to investigate how effectively inexpensive and widely available materials could be used to construct a functional combustion calorimeter. Rather than using specialized laboratory-grade instrumentation, much of the apparatus was assembled from scrap metal HVAC components, consumer-grade tubing, basic thermocouples, and low-cost pumps and scales. This introduced substantial engineering and measurement challenges, particularly in minimizing thermal losses and accurately quantifying fuel consumption.

The experiment also provided an opportunity to study practical limitations of real calorimeters. In an ideal calorimetric system, all heat released by combustion would be transferred directly into the working fluid and measured perfectly. In practice, however, thermal energy is inevitably lost through conduction, convection, radiation, and incomplete combustion. As a result, experimental measurements often underestimate theoretical values unless careful calibration and thermal isolation techniques are employed.

Using our apparatus, we measured the thermal energy transferred into flowing water during propane combustion and used this to estimate the higher calorific value (HCV) of propane. We then compared our experimentally measured value to accepted literature data and to previously published inexpensive calorimeter designs in order to evaluate the performance and limitations of our system.

Theory

In this project, we seek to quantify the enthalpy of combustion of propane using a hybrid Boys-flow calorimeter. The enthalpy of combustion is defined as the amount of energy released as heat when one mole of a substance undergoes complete combustion under standard atmospheric conditions. During combustion, chemical potential energy stored in the molecular bonds of propane is converted into thermal energy through an exothermic oxidation reaction.

The ideal combustion reaction for propane is$C_3H_8+5O_2 \longrightarrow 3CO_2 + 5H_2O$

Under ideal conditions, all of the chemical energy released by this reaction would be transferred into the surroundings as usable thermal energy. The purpose of the calorimeter is to capture as much of this thermal energy as possible and infer the heat released from measurable temperature changes in a working fluid.

Water is particularly useful for calorimetry because of its large specific heat capacity,

meaning that relatively large quantities of absorbed heat produce only moderate changes in temperature. In our apparatus, propane combustion heats copper tubing carrying flowing water. Thermal energy is transferred from the combustion gases into the copper through conduction and convection, and subsequently into the flowing water. Copper is used because of its exceptionally high thermal conductivity, allowing rapid heat transfer from the combustion region into the water stream.

The heat absorbed by the water is approximated by$$Q = m_{\mathrm{water}} c_{\mathrm{water}} \Delta T$$

where $m_{\mathrm{water}}$​ is the mass of water flowed through the apparatus and $\Delta T = T_f – T_i$ is the measured temperature increase of the water. Dividing this quantity by the mass of fuel consumed gives an experimental value for the heating value of propane:$$HCV_p = \frac{m_{\mathrm{water}}}{m_{\mathrm{fuel}}} c_{\mathrm{water}} (T_f-T_i)$$

This expression assumes that all heat released during combustion is transferred into the flowing water. In practice, however, significant heat losses occur through several mechanisms. Thermal energy is absorbed by the copper tubing itself, conducted into the stainless steel shroud, radiated into the surrounding environment, and carried away by hot exhaust gases escaping from the top of the apparatus. As a result, the experimentally measured heating value systematically underestimates the true thermodynamic value.

Another important assumption is complete combustion of the propane fuel. In oxygen-poor regions of the flame, incomplete combustion can occur, producing carbon monoxide or elemental carbon (soot) rather than fully oxidized carbon dioxide. These alternate reaction pathways release less energy than ideal combustion and further reduce the measured heating value.

Materials and Methods:

Materials List

• 4 foot stainless steel HVAC shroud (any sort of metal tube will do)
• Insulation (car exhaust pipe)
• 16m 1/4 inch copper tubing
• Fish tank water pump
• Thermocouples
• Digital scale
• Burner
• Propane cylinder (400g net weight)
• Graduated cylinder (1000ml)

Construction of the apparatus:

Turbulence is exceptionally important to effective heat transfer inside of our apparatus. Without some turbulent flow, heat from the bunsen burner would form an air column that would try to travel straight up. To combat this, we first developed some heat spreaders (shown above), consisting of decently sized meshes that would disrupt the flow of heat up the column and hopefully onto our coils to dissipate. In addition, placing the bunsen burner directly under the coil (as evidenced by the soot build up on the copper)rather than centered in the shroud effectively introduces turbulence at the very point that the heat is entering our column. We note that the addition of these mechanisms increased our recorded downstream temperature by more than 10 ºC, highlighting their success.

The effectiveness of this turbulence generation can be understood from convective heat transfer principles. In smooth laminar flow, hot gases tend to rise in relatively narrow vertical plumes, minimizing contact with the copper tubing and reducing overall thermal transfer efficiency. By introducing mesh heat spreaders and offsetting the burner beneath the coil, the upward flow of combustion gases became significantly more chaotic and spatially distributed. This increased mixing forces hot gases into greater contact with the copper surface area and improves convective heat transfer into the tubing.

The substantial increase in downstream water temperature following the addition of these turbulence-generating mechanisms suggests that convective transport, rather than conductive transport through stagnant air, dominated the thermal coupling between the flame and the copper coils. The observed increase of more than $10^\circ\mathrm{C}$ in downstream temperature demonstrates that airflow management was one of the most important factors affecting calorimeter performance.

Data

Raw Data

Run Number	Duration (min)	Ti: Reservoir Temperature (ºC)	Tf: Downstream Temperature (ºC)	Mass of water flowed (g)	Mass of propane flowed (g)**
1	10	21.8	41.6	3049	7.04
2	10	22.0	33.0	3139	2.35
3	10	22.0	41.4	3202	9.60
4	10	22.0	39.0	3155	8.07
5*	5	22.0	70.9	1633	15.98

We observe inconsistent, unreproducible results with the above methodology. Additionally, despite taring the scale before placing the fuel cylinder and beginning the experiment, upon completion of the trial and removal of the scale, it consistently read a positive, indicating drift had occurred. We attempt to characterize this by adding the drift mass into the mass of fuel consumed as read by the scale, because we believe the sum of these two quantities should be the total fuel mass consumed. The following data is shown below. We will proceed to only analyze and reference this corrected data for the remainder of this analysis

Corrected Data

Run Number	Duration (min)	Ti: Reservoir Temperature (ºC)	Tf: Downstream Temperature (ºC)	Mass of water flowed (g)	Scale drift mass (g)	Mass of propane flowed (g)**
1	10	21.8	41.6	3049	4.16	11.20
2	10	22.0	33.0	3139	7.15	9.50
3	10	22.0	41.4	3202	2.94	12.54
4	10	22.0	39.0	3155	3.55	11.62
5*	5	22.0	70.9	1633	1.60	17.58

Analysis

As described in the theory section, we can use the formula

to characterize the enthalpy of combustion of the propane fuel used, this is reported in the following table. Using this formula, and the specific heat of water (c_water) given by 4.184 kJ/kgºC, we determine experimental values of HCV_propane as shown in the following table. We note that

Run Number	HCVpropane (kJ)
1	22.5 * 103
2	15.2 * 103
3	20.7 * 103
4	19.3 * 103
5	19.0 * 103

We note that the dominant source of uncertainty in this experiment originated from measurement of the propane mass consumed. Temperature uncertainties were estimated to be approximately $\pm 1^\circ\mathrm{C}$, corresponding to roughly $5\%$ uncertainty in the measured temperature rise due to fluctuations in downstream temperature during operation. By contrast, the mass of water flowed through the system was measured using an electronic scale with uncertainty well below $0.1\%$, making its contribution negligible.

The fuel mass measurement, however, suffered from substantial systematic and random uncertainty. Despite taring the scale prior to each trial, the scale consistently exhibited drift over the duration of the experiment, likely caused by thermal effects from nearby combustion heating as well as internal instability in the inexpensive scale electronics. Because the observed drift magnitude was often comparable to the measured propane consumption itself, the uncertainty in fuel mass became the dominant limitation of the experiment.

To partially compensate for this issue, we introduced a “drift correction” by adding the observed post-trial scale offset to the measured fuel consumption. While this approach improved consistency between trials, it remains an imperfect correction because it assumes the drift occurred linearly and entirely as an offset rather than as dynamic fluctuations during operation. Consequently, we conservatively estimate the uncertainty in fuel mass to be on the order of $50\%$.

When uncertainties are combined in quadrature, the total uncertainty is overwhelmingly dominated by the fuel mass measurement. This yields a relative uncertainty of approximately $54\%$, resulting in the experimental value$$HCV_P = 19 \pm 10 \ \mathrm{MJ/kg}$$To better characterize the uncertainty in our measurements, we turn to sample statistics. In a fully efficient calorimeter, all of the heat of reaction would be redirected and recorded through the instrument, regardless of flame height, water flow rate, and any other factors, thus yielding the same HCV value. As such, to properly characterize the success of our instrument, we make the same assumption and assume independence of the above five trials allowing us to perform statistics on our results. This yields the following value for the 95% confidence interval of the enthalpy of combustion of propane:

Although the propagated uncertainty is extremely large, the spread between independent trials is notably smaller than the worst-case uncertainty estimate would suggest. This indicates that many systematic effects in the apparatus may have remained relatively stable between runs. The comparatively narrow statistical confidence interval therefore likely reflects the repeatability of the apparatus under similar operating conditions, rather than its absolute thermodynamic accuracy.

Nevertheless, the measured value remains substantially below the accepted literature value of $50.3 \ \mathrm{MJ/kg}$. This discrepancy strongly suggests that a large fraction of the combustion energy escaped the apparatus rather than being transferred into the flowing water. Likely contributors include radiative losses from the exposed copper tubing, conductive losses into the steel shroud, incomplete capture of hot exhaust gases, and incomplete combustion within oxygen-poor regions of the flame.

In practice, it is nearly impossible to make a fully efficient calorimeter, even more so when using cheap scrap metal in the construction of the apparatus. As such, to properly characterize the success of this experiment it makes more sense to compare to Quigley’s design which reports a relative error of 25%. As compared to our 62%, their value is significantly more accurate, indicating their experimental design was superior to ours. This discrepancy will be further discussed in the future work section.

Future Work

A large portion of our issues with accurately quantifying the enthalpy of combustion were as a result of our mass measurement systems. We attempted to use a cheap load cell, but had problems with the repeatable accuracy of the cell under almost a kilogram of the propane cylinder and torch head. We then attempted to use a digital scale, and subsequently had issues with thermal transfer into the scale, with cooling causing drift, as well as large amounts of scale drift as discussed in our data analysis. In the future, a more expensive load cell as used by Quigley et al. would likely be helpful in improving the accuracy of fuel mass measurements.

Another avenue to explore would be the burner types. Bunsen burners typically do not spread heat that far in the horizontal plane, and plume vertically. The burner used by Quigley et al. was more akin to a campfire stove burner, which spreads heat out in a circular pattern of large radius (similar to the coil curl radius). Our attempts to create turbulence through heat spreaders would have largely been moot if we had used such a burner instead.

The coil length in our apparatus was largely unnecessary. Given by the speed at which the water cooled once the burner was turned off and heat stopped being added to the system, only ~12-18 inches of vertical height of copper had any heat transference at all. Reducing the coil length would allow a lighter, more portable setup, without significantly impacting heat transfer. Quigley et al. only used 9 meters of coil to our 16 meters, and achieved still better efficiency.

Another important improvement would be the implementation of active thermal isolation and heat recovery systems. A significant amount of thermal energy visibly escaped from the top of the shroud during operation, indicating that much of the combustion heat bypassed the copper tubing entirely. Future designs could incorporate secondary heat exchangers, improved exhaust baffling, or reflective thermal insulation to redirect more heat back toward the coils. Additionally, enclosing the system more completely could reduce radiative and convective losses into the surrounding environment.

A further improvement would involve direct calibration of the calorimeter efficiency. Rather than assuming that all transferred heat is captured by the flowing water, the apparatus could first be tested using an electrical heater with a known power output. Comparing the measured thermal uptake of the water to the known electrical energy input would allow determination of an experimental calorimeter efficiency factor. This calibration could then be used to correct measured combustion values and substantially improve agreement with theoretical predictions.

References

National Institute of Standards and Technology. NIST Chemistry WebBook, SRD 69; Thermophysical Properties of Fluid Systems. https://webbook.nist.gov/cgi/cbook.cgi?ID=C74986&Mask=1 (accessed May 11, 2026).

Quigley G, Cumber P, Houston S. A novel inexpensive calorimeter as a thermodynamic laboratory experiment. International Journal of Mechanical Engineering Education. 2026;54(2):391-409. doi:10.1177/03064190241303824